Investigation of Frequency Analysis Methods for Doppler Ultrasound Systems

  •  Tatsuro Baba    


Due to the advances of electronic and semiconductor technologies in recent years, it is possible to realize complex, low cost, low size, and low power consumption, high-speed signal processing devices. The progress of these devices has enabled the development of the medical Doppler ultrasound system. Color flow mapping (CFM), which is one of the display mode of Doppler ultrasound, requires high-speed multi-point (two- or three-dimensional) frequency analyses. From its birth till today, a complex autocorrelation (AC) method has been used for CFM because of its simplicity. In this paper, I propose the fast Fourier transform (FFT) method for the frequency analysis of CFM. CFM differs from spectrum Doppler, which shows accurate information of the blood flow in a narrow domain of a tomogram image. CFM uses color expression to display coarse information of the blood flow, such as mean velocity, intensity, and distribution. Because the calculation load of the frequency analysis is very small, the AC method has been used. However, by exploiting recent advances in hardware, new frequency analysis methods can be applied. In this paper, I evaluate a novel frequency analysis method based on FFTs, and compare its performance with the conventional AC method. Based on the results obtained, I reach the followings conclusions. With respect to mean velocity, the FFT method performs well when blood flow sensitivity is low. However, when blood flow sensitivity is high, the performance of the AC method is superior. Moreover, with respect to the distribution, compared to the FFT method, the AC method does not perform well under aliasing conditions. The AC method is effective only when the distribution is small.

This work is licensed under a Creative Commons Attribution 4.0 License.
  • ISSN(Print): 1916-9639
  • ISSN(Online): 1916-9647
  • Started: 2009
  • Frequency: semiannual

Journal Metrics

Google-based Impact Factor (2017): 3.90
h-index (November 2017): 17
i10-index (November 2017): 33
h5-index (November 2017): 12
h5-median (November 2017): 19

Learn more